Sarah Alnegheimish research interests are located at the intersection of machine learning and system engineering. Her goal: Making machine learning systems more accessible, crystal clear and trustworthy.
Early influence
The daughter of a university professor and teacher learned from an early age that knowledge was to be freely divided. “I think that growing up in a house where education was highly valued is part of why I want to share machine learning tools.” The personal experience of Alnegheimish with Open Source resources only increased its motivation. “I have learned to perceive availability as the key to adoption. To strive for influence, you should access the new technology and evaluate by those who need it. This is the purpose of developing open source.”
Alnegheimish obtained a bachelor’s degree at King Saud University (KSU). “I was in the first cohort of computer science. Before creating this program, the only available direction in the calculations was [information technology]. “Being part of the first cohort was invigorating, but it brought its unique challenges. Successful required independent educational experience. Then for the first time I came across the myth of OpenCurseware: as a resource to learn.”
Shortly after graduating, Alnegheimish became a researcher at King Abdulaziz City for Science and Technology (Kacst), the National Saudi Arabian Laboratory. Through the Center for Complex Engineering Systems (CCE) in KACST and MIT, she began to conduct research with Veeramachannia. When she applied to the myth to the postgraduate school, his research group was her best choice.
Creating Orion
Orion uses statistical models and machine learning, which are constantly registered and maintained. Users do not have to be machine learning experts to use the code. They can analyze signals, compare anomalies detecting methods and study anomalies in a comprehensive program. The frames, code and data sets are open.
“Thanks to Open Source, availability and transparency are directly achieved. You have unlimited access to the code in which you can examine how the model works by understanding the code. We have increased transparency with Orion: we mark each step in the model and present it to the user.” Alnegheimish says that this transparency helps users start trusting the model before they finally see how reliable it is.
“We try to take all these machine learning algorithms and put them in one place so that everyone can utilize our models on the shelf,” he says. “It’s not just about the sponsors with whom we work in the myth. It is used by many public users. They come to the library, install it and run their data. It turns out that this is a great source for people to find some of the latest methods of detecting anomalies.”
Changing models for detecting anomalies
Available project
Alnegheimish talks for a long time about the efforts she has gone to make Orion more available. “Before I came to the myth, I thought that the key part of the research was the development of the machine learning model itself or improving its current state. With time I realized that the only way you can make and adapt to research for others is to develop systems that provide them. During post -graduate studies I took the approach to developing my models and tandem systems.”
The key element of its system development was finding the right abstractions to work with its models. These abstractions provide universal representation for all models with simplified components. “Each model will have a sequence of steps to go from a strict entrance to the desired exit. We standardized input and output, which allows the flexibility and fluid of the center. So far, all the models we launched have been able to modernize our abstractions.” The abstractions he uses have been stable and reliable over the last six years.
The value of simultaneous building of systems and models can be seen in the work of Alnegheimish as a mentor. She had the opportunity to cooperate with two students of the master earning in engineering. “I only showed them the system and documentation how to utilize it. Both students were able to develop their own models with abstractions with which we adapt with them. It confirmed that we are going the right path.”
Alnegheimish also examined whether a large language model (LLM) can be used as a mediator between users and the system. The LLM agent who implemented is able to connect with Orion without users who will have to learn the small details of Orion. “Think of chatgpt. You have no idea what is behind him, but it is very available to everyone.” In the case of its software, users know only two commands: match and detect. Fit allows users to train their model, and Detect allows them to detect anomalies.
“The ultimate goal of what I tried to do is to make AI more accessible to everyone,” he says. Until now, Orion has achieved over 120,000 downloads, and over a thousand users marked the repository as one of their favorite in Github. “Traditionally, you have an impact of research through paper quotes and publications. Now you get real -time adoption through Open Source.”